NVIDIA GeForce GTX 1070 On Linux: Testing With OpenGL, OpenCL, CUDA & Vulkan

[jf33]

Even with Tesseract, the R9 Fury doesn't come close to competing with the GTX 1070.

What a stupid comment. Of course it doesn't! It was never supposed to. Even in D3D 11 games, the Fury is only about as fast as a GTX 980.

[humbug]

Something is really wrong with AMD Vulkan driver or DOTA2 Vulkan renderer.

Michael also noted in a previous article that the dota2 vulkan renderer is under-performing at GPU bound high resolutions similar to what you posted
https://www.phoronix.com/scan.php?pa...an-redux&num=4

Whereas at 1080p and 1440p resolutions it's doing well
https://www.phoronix.com/scan.php?pa...an-redux&num=3

[bridgman]

Yeah, something is definitely wrong there. In the "Steam Client Stable..." article a non-X Fury ran at 81.23 fps at 4K with the overlay fix (59.10 without), while in this article the old 59.10 number appeared again. The 290 ran at 63.7 fps in the other article but 12.27 here

I do have to gently protest that AMD hardware seems to have been excluded from most of the tests where it would have performed well, eg Luxmark scores are not displayed although Luxmark perf-per-watt numbers are, suggesting that the tests were actually run on AMD hardware.

[bridgman]

What a stupid comment. Of course it doesn't! It was never supposed to. Even in D3D 11 games, the Fury is only about as fast as a GTX 980.

And yet in DX12 games it seems to outrun the 980ti and compete with the newer chips. Go figure.

[Michael]

Yeah, something is definitely wrong there. In the "Steam Client Stable..." article a non-X Fury ran at 81.23 fps at 4K with the overlay fix (59.10 without), while in this article the old 59.10 number appeared again. The 290 ran at 63.7 fps in the other article but 12.27 here

I do have to gently protest that AMD hardware seems to have been excluded from most of the tests where it would have performed well, eg Luxmark scores are not displayed although Luxmark perf-per-watt numbers are, suggesting that the tests were actually run on AMD hardware.

Luxmark on AMDGPU-PRO crashed for the scenes used in my main article but where I was doing the performance-per-Watt run in a separate result file it happened to be with a LuxMark scene that worked on AMDGPU-PRO.

[rabcor]

Why are the cards so physically big though? I mean in the early days of boasting about pascal's feats one of the key boasts that their CEO made was that the cards were only about the size of two credit cards... this is clearly a lot bigger than that on any axis. More like 10 credit cards, or one maxwell gpu.

[bridgman]

Luxmark on AMDGPU-PRO crashed for the scenes used in my main article but where I was doing the performance-per-Watt run in a separate result file it happened to be with a LuxMark scene that worked on AMDGPU-PRO.

Interesting... so sounds like it crashes on Hotel but runs well on Microphone and Luxball HDR ?

Just curious, why different scenes used for performance and performance/watt ?

[computerquip]

With the demanding Unigine Valley, the GTX 1070 even edged out slightly ahead of the GTX 1070 for greater performance-per-Watt.

You know the card is good when it's better than itself.

[bridgman]

maybe they rollback the update in stable dota2 because some people had regressions with the update?
I read about problems like this in the phoronix news forum threat the patch originally showed up.

AFAIK the change improving overlay performance was in the Steam client not DOTA2 itself, and there's no sign of a Steam client rollback I can see.

There have been a couple of DOTA2 updates but it's incredibly unlikely that they would cause a performance regression matching what the Steam client fixed to four significant digits

[BillBroadley]

My theory on video cards (and hardware in general) is generally:
* buy better than you need today and keep it for longer
* 10-50% is not really noticeable, wait for a 2x performance improvement
* ignore performance on things not performance limited (like games with 200+ fps)
* improve worst case performance when you can (ssd, ram, extra memory bandwidth, or extra ram)

With that in mind, the GTX 970 does pretty well against the 1070. For the games with reasonable frame rates under 150 hz average, where you might occasionally notice lag as it drops under 60 fps, the 1070 is 15-35% faster. Nice for sure, but I'd wait a generation if you have a 970. Now if you have a gtx 760 or slower it's a pretty big win.